Claiming R&D on Computer Vision
Computer vision has been around for more than half a century and it is one of the most popular fields of artificial intelligence (AI). Despite tremendous advances, there are still plenty of challenges the field aims to overcome. This, as there is still a gap in understanding the intricacies of the way human receptors like the eye work as well as how the brain processes visual stimuli. There are also limitations in the tools and technologies up to this day— for instance, in hardware used for detection such as cameras or sensors, or in data volume or quality. As such, people working on computer vision can make use of the Research and Development Tax Incentive scheme to support the growth of the industry.
Computer vision began by merely detecting edges and corners to determine the basic shapes of objects. The growth of AI, deep learning, and the advent of more advanced algorithms such as convolutional neural networks (CNN), paved the way for huge advancements in the industry. Now, computer vision has many groundbreaking applications. For instance, it is being continuously developed to detect early signs of cancer and drive cars autonomously, among others.
Below are examples where computer vision is heavily applied:
- Image classification and segmentation
- Facial recognition
- Fingerprint recognition and biometrics
- Optical character recognition
- Automotive safety
- Defect detection
- Medical imaging
When claiming computer vision activities, it is essential to note that AusIndustry looks for specific activities conducted within the development of a project, not the project in its entirety. Let’s say, a company builds new scanning software that uses computer vision, and the product does not exist yet on the market. If the product merely uses an already existing computer vision algorithm, it will be difficult to make it qualifiable R&D. The company must prove the uniqueness of a specific process undertaken or a technique developed when using computer vision.
Examples of unique processes or techniques include:
- Developing a new feature detector
- Creating a new segmentation technique
- Writing a new algorithm to improve the accuracy of text scanning with low-quality physical images
An R&D activity is composed of the main activity which is called a core activity, and activities that directly support it are aptly called supporting activities. To claim core activities, companies must be able to show that they have conducted a series of experiments with the purpose of solving technical, not business challenges, as well as to produce new knowledge in the field.
We at Innercode can help you assess whether your computer vision activities are eligible as R&D activities for the R&D Tax Incentive. Through our streamlined process of analysing claims, we can help you determine which activities we should focus on to maximise your R&D tax claim with just a few hours of your time. We have years of experience delivering the R&D Tax incentive and other government grants to various software and tech companies– many of which are building enterprise and consumer computer vision-related projects.